Named Entity Recognition on MasakhaNER
59.4Macro F1Gemma3-27B
Evaluation results
| Method | Links | |
|---|---|---|
| Gemma3-27B2026.01 | 59.4 | |
| OTTERArchitecture=Cross-Encoder, Backbone=RemBERT2026.01 | 57.3 | |
| GLiNER-x-base2026.01 | 55.9 | |
| OTTERArchitecture=Bi-Encoder, Backbone=mmBERT2026.01 | 55.3 | |
| OTTERArchitecture=Bi-Encoder, Backbone=RemBERT2026.01 | 54.1 | |
| Qwen3-32B2026.01 | 53.5 | |
| OTTERArchitecture=Cross-Encoder, Backbone=mmBERT, Training Steps=100k, Thresholding Strategy=Best per language2026.01 | 52.3 | |
| OTTERArchitecture=Cross-Encoder, Backbone=mmBERT2026.01 | 51.1 | |
| OTTERArchitecture=Cross-Encoder, Backbone=mmBERT, Training Steps=100k2026.01 | 51.1 | |
| OTTERArchitecture=Bi-Encoder, Backbone=mmBERT, Training Steps=100k2026.01 | 48.1 | |
| GLiNER-multi-v2.12026.01 | 48 | |
| GPT-52026.01 | 38.8 | |
| WikiNeural2026.01 | 30.7 |